GlueX-doc-1612SVN: docs/GlueXdoc/short cover

Original: September 24, 2010

Overview of Hall D solenoidrefurbishment and testing

G. Biallas, E. Chudakov, E.S. Smith, T. Whitlatch, E. Wolin

1 Overview

The solenoid magnet provides a 2.2 T magnetic field parallel to the beam directionat the center of the magnet. The inside diameter of the magnet is 185 cm and itsoverall length is 465 cm. In the configuration for Hall D, the inductance of thecoil is 26.2 H, and the magnet will run at the nominal current of 1500 A resultingin a stored energy of 29.5 MJ. The iron and coil configuration is shown in Fig. 1with the calculated magnetic field lines superimposed. The magnet was designedand built using standards that today would be considered very conservative. It is acryostatically stable design and uses cryostats that were designed to be opened andserviced with hand tools. The magnet in its original configuration and use at SLACare described in the technical note of Ref. [1]. Table 1 summarizes the importantmagnet parameters.

The solenoid is constructed of four separate superconducting toroidal coils andcryostats. A 306 ton iron flux return path, similarly split in four places surroundsand supports the coil assemblies. A common liquid helium reservoir is located atopthe solenoid providing the gravity feed of the liquid to the coils. Full access to theinterior is provided from both ends of the magnet for installation and maintenance ofdetectors.

The superconductor is a stabilized, composite, twisted multi–filament wire ofniobium–titanium. The wire is made by soldering the superconductor compositebetween two copper strips to form a rectangular cross section 0.763 x 0.533 cm2 [2].The coil was wound directly on the cylindrical inner wall of the liquid helium ves-sel. As the coil was wound, a 0.025 inch-thick stainless steel support band and two0.0075 inch-thick Mylar insulating strips were wound along with it for mechanicalsupport and insulation. Cooling by the liquid helium is accomplished from the edges.Each of the four toroidal windings has its own predetermined number of turns, theconfiguration originally being selected for optimum central field homogeneity.

Each liquid helium vessel is surrounded by a liquid nitrogen cooled radiation shieldand this assembly is centered in the vacuum tank by a circumferential series of tiebolts designed for minimum conductive heat flux to the helium cryostat. Radialcentering and support are achieved by adjusting screws built into the four tie bolts.

1

Characteristic Value

Central field 2.2 TOperating current 1500 ATotal stored energy 29.5 MJInductance 26.2 H

Inside diameter of coils 203 cmClear bore diameter 185 cmOverall length (iron) 495 cmOutside iron diameter 376 cmTotal iron weight 306 TTotal magnet weight 350 T

Max force on axial tie bolts due to magnetic forces 7500 lb(coil 2 at 1100 A) (25,000 lb rating)

Total Helium volume (including reservoir) 4,000 lHelium heat load (refrigeration) 38 WHelium system electrical lead flow (liquefaction) 0.2 g/sStatic nitrogen load 30 l/hrLiquid helium storage dewar 4000 lLiquid nitrogen storage dewar 10000 lCool-down time 3 weeks

Ratio of copper to superconductor 20:1 (Grade A)in composite 28:1 (Grade B)Total conductor length 35.8 kmTotal conductor weight 13.2 T

Turn on time (normal) 1.5 hTurn off time (fast dump) 20 minDump resistor 0.072 ΩNumber of turns 4608

Table 1: Summary of characteristics of the solenoid in the Hall D configuration.

2

GlueX (Hall D) solenoid − basic, 1500A

Z:\home\gen\.wine\my_poisson\GLUEX_SOL\GLUEX_SOL_04.AM 5−12−2010 21:03:52

0

50

100

150

200

250

300

350

0

50

100

150

200

250

300

350

−50 0 50 100 150 200 250 300 350 400

Figure 1: Magnetic field for the Hall D configuration at 1500 A computed withPoisson.

Selected tie bolts are instrumented with strain gauges to measure the forces on theassembly caused by magnetic fields.

The liquid helium volume is about 5000 liters and the heat load is estimated at60 liters/hr equivalent. The volume of LHe for each coil is as follows: 1150 l (coil 1),560 l (coil 2), 610 l (coil 3), and 460 l (coil 4). Refrigeration is provided by a CTICryogenics Model 2800 4.5 K helium refrigerator, which can deliver a linear mix ofup to 200 W refrigeration or 2 g/s liquefaction. This conservative design, both of therefrigeration system and of a superconductor itself, has produced a stable, reliableand safe superconducting magnet.

2 Brief history

The superconducting solenoid was built in the early 1970’s for the LASS detector atSLAC. It operated at SLAC for experiment E-135 in 1981 and 1982 (see for exampleRef. [3] and references therein). The experiment used all four coils, labeled as 1–4, adesignation that we continue to use for Hall D even though we will use a configurationwith a different order for the coils. The nominal central field for LASS was 2.24 Tfor a current of 1600 A1, although the magnet was originally designed to operate upto 1800 A.

1Steve St. Lorant, original designer and magnet reviewer recalled that the magnet was not runabove 1250 A, but the literature specifies operation at 1600A.

3

The solenoid was moved to LAMPF at Los Alamos National Laboratory (LANL)in 1985, where it was used for the MEGA experiment. The production data forMEGA was collected between 1993 and 1995 (see for example Ref. [4] and referencestherein). This experiment only used coils 1–3 and coil 4 was kept in storage until thepresent refurbishment effort for Hall D. The central field was 1.5 T operating at acurrent of 1178 A.

The solenoid was inspected in April 2000 by a team consisting of JLab staff andtwo of the original designers of the magnet, John Alcorn and Steve St. Lorant.The team concluded that the coils had been kept in excellent condition and decidedto transfer the magnet to Jefferson Lab for use in the GlueX experiment in HallD at Jefferson Lab. The magnet was transfered to the Indiana University CyclotronFacility (IUCF) in October 2002 to refurbish the coils in preparation for use in GlueX.Upon completing work on coils 1, 2 and 4, these were transfered to JLab in summerof 2005, where the refurbishment effort continues. The yoke rings and iron end pieceswere moved to JLab in October of 2006. The final closure of coil 3 is presentlyunderway at IUCF.

3 Modifications and repairs

We briefly describe the changes and repairs that are being made to the magnet forits use in Hall D. These include changes to a) the configuration of coils and yoke, b)refurbishment, c) repairs of shorts to ground, d) new cryogenic system and controls,and e) compliance with new pressure safety guidelines.

3.1 Configuration of coils and iron yoke

There are several changes to the configuration of the coils and iron yoke which havebeen implemented to the magnet for use in Hall D. The major changes are the fol-lowing:

1. enlarging of the upstream mirror plate to a diameter of 185 cm for ease ofaccessing detectors inside the bore,

2. switching the positions of coils 1 and 2, resulting in a coil configuration fromupstream to downstream of (2,1,3,4). This change together with the additionof 3” iron baffles between coils 2 and 1, 1 and 3, and 4 and the downstreamyoke reduces the axial forces on the coils.

3. adding iron at strategic locations to reduce saturation of the iron and reducethe fringe field of the magnet. The added iron includes cladding on the exteriorsurface of the yoke covering about 75% of the area, and increasing the lengthof the downstream iron ring by 6”.

4

3.2 Refurbishment

The refurbishment effort consisted of repairs and maintenance that did not requireopening the helium vessel or work directly on the coils. It included finding and re-pairing the numerous leaks that have plagued the solenoid during previous operation.Coils 1, 3 and 4 had leaks in the nitrogen circuits and coil 2 had a leak in the LiquidHelium vessel. All of these were repaired at IUCF, with coil 3 requiring the replace-ment of its nitrogen shield. After repairs, all circuits were checked to have leak ratesbelow 10−8 torr-liter/s.

The shield thermometry, which originally was based on thermocouples, was re-placed with more accurate and modern PT-102 Platinum resistance thermometers,which are now the industry standard. We note that the carbon resistor thermometersinside the helium vessels, to check LHe levels, have been checked and retained. Thecoil support strain gauges were all replaced as many of the originals had failed overthe years. Finally the multi-layer super insulation, which had deteriorated from pre-vious maintenance and operations with oil diffusion pumps was replaced. During thecourse of this maintenance work, corrosion was discovered in some internal plumbingand all four 18-inch coil junction bellows. These have all been replaced.

3.3 Shorts to ground and repairs inside the helium vessels

Repairs inside the helium vessel were necessary to fix shorts to ground and reinforcethe supports of sub-coils within coil 2. The first short to ground in the magnetdeveloped in coil 4 at SLAC in 1973. This short was traced to a failure of the groundinsulation between the titanium springs and the stainless axial straps and betweenthe stainless straps and ground. Coil 4 was fixed by opening it and inserting G-10strips between the coil and the stainless straps. After the repair all four coils passedhi-pot tests.

Shorts to ground were noted in 1986 in coil 3 (∼ 0.1 Ω) and in coil 1 (∼ 25 Ω), andwere presumably present during the operation of the solenoid for the MEGA exper-iment. After arrival at IUCF all coils were re-measured. Coil 3 confirmed the samehard short to ground, and coil 1 had an observed short of 1.85 Ω to ground. Bothshorts were found at the same locations as the ones measured at Los Alamos[5].2 Dur-ing repairs it was also determined that these shorts were due to a similar mechanismas that of the first short found in coil 4 and fixed at SLAC years earlier. There aretwo locations in each coil where such shorts to ground can develop, for a total of eightin all coils. Four locations have been shored up, corresponding to the downstreamend in each coil. The short in coil 3 was fixed at IUCF and the short in coil 1 wasfixed at JLab.

2A logbook page from 1975 indicates that a 25 Ω short in coil 1 was observed at SLAC. Thelocation seems to be different from the short found in coil 1 but this may simply be a case ofconfusion over different definitions of coil polarity.

5

One final operation was needed that requires opening one of the helium vessels.The force analysis for the new Hall D coil configuration found that sub-coils withincoil 2 lacked adequate support. Therefore, additional buttressing is being added tothis coil to solidify the structural support, which is an ongoing task at JLab. Thehelium vessel was opened, a repair strategy has been devised and the repair is waitingfor supports to be fabricated. Upon exposing the coil, it was found that the steel striphad shorted to ground, but it has now been isolated. Therefore, there are no longerany shorts to ground in any of the four coils.

3.4 Cryogenic supply and control system

The cryogenic interface and control system for the magnet is completely new andbased on proven designs currently in use on seven different SC magnets at JLab. TheJLab cryogenic supply system delivers liquid nitrogen (LN2) at 77K and 3 psi, andsupercritical He at 4.5K and 3 atm. Liquid He (LHe) is made in situ via expansionof the supercritical He through a Joule-Thompson valve.

The solenoid cryogenic system consists of a “cryo can,” containing a Joule-Thompsonvalve and small internal LN2 and LHe reservoirs, sitting on top of a “cryo box” con-taining large internal LN2 and LHe reservoirs. The external cryogen supply andreturn lines connect to the cryo can, which then feeds the cryo box and receivesreturn gases from the box. LHe is made in the cryo can. The box distributes thecryogens to the four coils, receives the return gases from the coils and feeds them tothe cryo can for return to the JLab cryogenic supply system.

The control system is based on Allen-Bradley PLC’s, chosen due to the extensiveexperience with this PLC line at JLab. This modern control system replaces theoriginal manual system used for LASS and MEGA, allowing for PLC-based feedbackcontrol loops. This eliminates the need for very large cooling capacities and reservoirswhich were necessary to accommodate short-term fluctuations in operating conditionswhen the system was under manual control. Most of the cryogenic system is commonto all four coils. Specific to each coil are valves in the cryo box that control LN2 andLHe flow from the reservoirs to the individual coil cryostats, and temperature, strainand voltage sensors within each coil. We note that hard-wired interlock systems pro-tect all relevant components from operating outside their allowed ranges, independentof PLC control.

The solenoid is powered by a new Danfysik System 8000 Type 854 10V DC powersupply and energy dump system. The supply was purchased at the same time as anidentical unit for Hall C, which has run successfully for a number of years and willcontinue to be used for 12 GeV operations. We are working with Hall C to developa joint strategy on spare parts, maintenance and repair.

6

3.5 Compliance with pressure safety guidelines

The Pressure System of the Solenoid includes both Liquid Helium and Liquid Nitro-gen Circuits made of ductile stainless steel and copper. They are considered LegacySystems built pre 10CFR851 implementation at JLab and enjoy some relaxed em-phasis on material and weld pedigree. The nitrogen shield and helium piping designis robust for any overpressure scenario. However, the toroidal shaped coil vessels aremade with thin flat heads and non-standard weld joints that limit rating to 30 psia.Rare failure scenarios could raise pressures in the helium vessel during an event suchthat rupture is possible. We reduce the Probability of the most likely scenario –loss ofinsulating vacuum– by armoring thin walls and using engineered controls on the vac-uum valve. We reduce the Consequence Level of a failure by redefining the pressureboundary to the robust outer wall of this double wall vessel system and assuring thatthe outer wall is very well vented. Reduction in both Probability and ConsequenceLevel reduces the risk code (ESH&Q 3133) to Negligible. This low code shows wefulfilled the mandate of 10CFR851 by making the system equivalently safe to theASME Consensus codes.

4 Turn-to-turn shorts

During the process of fixing the shorts to ground, the helium cryostats of coils 1, 2and 3 were opened and their condition was examined. Several of the copper conduc-tors were found to be shorted to the supporting stainless steel strip, although thelocation of the contacts could not be identified. In addition, metal chips have beenfound inside the cryostats and such fragments could bypass the Mylar insulation onboth sides of the steel strip to create a short between coil windings. A visible shortbetween windings was found upon opening coil 1. However, inspection of this areasuggests that it did not occur at operating temperatures [6]. This particular shortwas accessible and has been repaired.

Nevertheless, there remains a potential for shorts between turns inside the coilwindings and their consequences need to be evaluated, especially during a quench.The report of the Lehman Review in April 2010 [7] recommended to “fully assess therisk of additional shorts, [and a] quantitative analysis of the impact of shorts andresulting over-currents and high voltages should be extended to worst case situations,such as multiple-turn shorts and low-resistance shorts.” Two possible scenarios ofshorts were considered:

1. The Mylar insulation between the copper cable and the stainless steel band isbroken on both sides of the band. In this case, the short is through the stainlesssteel band, and the worst case occurs when the contacts are directly across theband, which creates a short with a resistance at 4 K of 3.1 µΩ. A short of thistype is our main case study.

7

Parameter Symbol Nominal Value Range Studied

Solenoid inductance LS 26.2 H 26.2 HTurn inductance LT 10 µH 9-13 µ HMutual inductance κ

√LS LT 5.5 mH 4.0 – 8.1 mH

Mutual coupling κ 0.34 0.25–0.50Dump resistor RD 0.072 Ω 0.06–0.072 ΩResistance of short RS 3.1 µΩ 10 nΩ – 100 µΩResistance of one turn RT (superconducting) 0 Ω 0 Ω(resistive) 2.9 mΩ (300 K) 15–77 µΩ (4 K, RRR=38-200)Critical current Ic 4 kA 1–13 kACooling effectiveness AC 0.5 W/cm-K (∆T=1) 0.1 – 0.5

Table 2: Parameters used in the turn-to-turn short analysis in the solenoid. Wegive their nominal values and the range of values that were used in evaluating theconditions at the time of a quench.

2. A copper chip with a 100 mm2 cross section by-passes the stainless steel bandand two Mylar strips creating a short. At 4 K, this leads to a resistance as lowas 10 nΩ.

Clearly a wide range of conditions must be considered, and we give the rangeinvestigated in Table 2. We believe this set covers the limits of possibilities that couldoccur within the magnet. The response of the system involves an interplay betweenthe currents induced in the electrical circuits and the cooling provided by the heliumbath. The current through the short heats the adjacent area, which changes theproperties of the conductor. The cable around may lose superconductivity dependingon the temperature and the current. The change of the cable resistivity affects thecurrents distribution between the shorted turn and the rest of the solenoid.

Two independent approaches were used to simulate the response of the system toa fast ramp down of the magnet. The first, which we summarize here, is a numericalsimulation of a full model of the electrical circuits and heat propagation in the coils[8]. The second made the simplifying assumption of adiabatic equilibrium for theheat transfer, and was used as a check on the full simulation [9]. Both calculationsare in good agreement.

We outline the algorithm of the full numerical simulation: The cable was splitinto 2 cm long sectors. It was demonstrated that a finer splitting would not improvethe accuracy. At each time step the following processes were simulated:

1. Modifications of the material parameters in each spacial element accordingly tothe local temperatures. The cable element may switch between resistive andsuperconductive states depending on the local temperature, the current and themagnetic field.

8

2. The current distribution in the coils accordingly to the updated resistances.

3. The heat generated by the electric currents.

4. The heat flow between the adjacent elements including the flow through thestainless strips to the adjacent turns.

5. The helium cooling of two sides of the cable.

6. Updating the temperatures of all the elements at the end of the step.

The size of the time step was optimized to provide calculation stability. The calculatedmaximum temperature, the induced currents, estimated resistance of the shorted turnand the power dissipated are plotted in Figures 2-3 for two different time scales.The solutions often show oscillations of the current through the short and of othervariables. They result from two processes with their own time scales - the currentpumping into the shortened turn, and heating/cooling of the cable involved. For themaximum calculated power deposited, about 10% of helium may evaporate from thehelium vessel.

Within the range of our input assumptions, we conclude that that the maximumtemperatures in the magnet could possibly reach 60 K, but that the cryogenic systemshould be able to control a quench under these conditions. In summary, no dangerousoverheating was observed over the very wide spectrum of parameters used in thesecalculations.

5 Single-coil testing

The purpose of testing magnet coils individually is to verify coil cool-down calcula-tions, check coil operation, and to check that stresses observed when the coils areenergized agree with Finite Element model calculations. Each coil will be tested sep-arately with a reduced yoke, consisting of the iron ring for that coil and the iron endpieces. Judicious placement of the iron pieces relative to the coil will create stressesin the coil that mimic those expected in the full magnet. The calculated inductancefor the single-coil and reduced yoke configuration is 6.2 H The dump resistor is beingcorrespondingly reduced for the test to 0.022 Ω to approximately match the samevoltage drop per turn as in the full magnet. Testing will be performed at the operat-ing temperature of 4.2 K and operating current of 1500 A. All strain gauges, pressureand temperature sensors will be carefully monitored for anomalous readings.

The entire solenoid safety system will be tested during the single-coil tests. Hu-man and equipment safety systems for the test are currently under review by theJLab EHS&Q and 12GeV Safety groups, and they have approved our safety strategyduring initial discussions. Most of the cryogenic system will also be tested. PLCprogramming for the test is complete and ready for commissioning. For the entire

9

solenoid, the small number of additional sensors and actuators in the cryo box haveto be added and incorporated into the control algorithm, along with integration ofall four coils. Additional benefits of the test include training of installation crewsand operators, early construction of rigging fixtures and development of installationprocedures, and early education about and experience with solenoid operation. With-out the coil tests all this would have been done for the first time in the hall duringcommissioning of the complete solenoid magnet.

References

[1] D. Aston et al. The LASS spectrometer. SLAC-R-298.http://www.slac.stanford.edu/pubs/slacreports/slac-r-298.html, 1986. 1

[2] J. Alcorn. Actual Conductor Configuration and performance (Leith Solenoid).Log Book, June 2, 1972. 1

[3] D. Aston et al. Spectroscopy from the lass spectrometer. Nuclear Physics B -Proceedings Supplements, 8:32 – 39, 1989. 3

[4] M. Ahmed et al. Search for the lepton-family-number nonconserving decay µ+ →e+γ. Phys. Rev. D, 65(11):112002, Jun 2002. 4

[5] P. Brindza. Tests and calculations to determine the ground short characteristics ofthe gluex solenoid coils 1 and 3 and a discussion of consequences and remediationoptions. GlueX-doc-429, February 2005. 5

[6] P. Brindza, S. St. Lorant, and W. Schneider (chair). Report, Solenoid MagnetReporting and Planning Meeting. May 13, 2010. 7

[7] Department of Energy Review Committee Report. Technical, Cost, Schedule, andManagement Review of the 12 GeV CEBAF UPGRADE PROJECT. April 27-28,2010. 7

[8] E. Chudakov and E.S. Smith. Solenoid: Impact of a shortened turn. GlueX-doc-1610, September 2010. 8

[9] E. Chudakov and E.S. Smith. Currents and temperatures induced by turn-to-turnshorts during a quench. GlueX-doc-1611, September 2010. 8

10

20

40

60

0 5 10 15 20 25 30 35 40 45 50Time, s

TM

AX, K RRR=200, turn=4 RSHORT=0.20µΩ 1.00µΩ 5.00µΩ 10.0µΩ (a)

0

2000

4000

6000

8000

10000

0 5 10 15 20 25 30 35 40 45 50Time, sC

urre

nt in

the

shor

t, A

RRR=200, turn=4 RSHORT=0.20µΩ 1.00µΩ 5.00µΩ 10.0µΩ (b)

0

10

20

30

0 5 10 15 20 25 30 35 40 45 50Time, s

Sho

rted

turn

R, µ

Ω RRR=200, turn=4 RSHORT=0.20µΩ 1.00µΩ 5.00µΩ 10.0µΩ (c)

0

200

400

0 5 10 15 20 25 30 35 40 45 50Time, s

Pow

er to

hel

ium

, W RRR=200, turn=4 RSHORT=0.20µΩ 1.00µΩ 5.00µΩ 10.0µΩ (d)

Figure 2: The results of calculations of the processes in a shortened turn during thefast ramp-down through the 0.07 Ω dump resistor, shown in a time interval of 0-50 s.The calculations were made for RRR=200, a 0.01 T residual magnetic field, for aturn with a 12 µH inductance and the κ-parameter of 0.45. Plot (a) presents themaximal temperature in the cable, plot (b) presents the current through the short,plot (c) presents the resistance of the shortened turn, and plot (d) presents the powerabsorbed by helium.

11

20

40

60

100 200 300 400 500 600Time, s

TM

AX, K RRR=200, turn=4 RSHORT=0.20µΩ 1.00µΩ 5.00µΩ 10.0µΩ (a)

0

2000

4000

6000

8000

10000

100 200 300 400 500 600Time, sC

urre

nt in

the

shor

t, A

RRR=200, turn=4 RSHORT=0.20µΩ 1.00µΩ 5.00µΩ 10.0µΩ (b)

0

10

20

30

100 200 300 400 500 600Time, s

Sho

rted

turn

R, µ

Ω RRR=200, turn=4 RSHORT=0.20µΩ 1.00µΩ 5.00µΩ 10.0µΩ (c)

0

200

400

100 200 300 400 500 600Time, s

Pow

er to

hel

ium

, W RRR=200, turn=4 RSHORT=0.20µΩ 1.00µΩ 5.00µΩ 10.0µΩ (d)

Figure 3: The results of calculations of the processes in a shortened turn during thefast ramp-down through the 0.07 Ω dump resistor, shown in a time interval of 50-600 s. The calculations were made for RRR=200, a 0.01 T residual magnetic field,for a turn with a 12 µH inductance and the κ-parameter of 0.45. Plot (a) presents themaximal temperature in the cable, plot (b) presents the current through the short,plot (c) presents the resistance of the shortened turn, and plot (d) presents the powerabsorbed by helium.

12